How accurate are energy demand forecasts? I.e., what is the error of the estimate for forecasting future global energy demand, or future oil demand, a few years into the future? This data-file aggregates data in order to answer this question, across oil, coal, gas and renewables. The best rule-of-thumb is 0.6% per year.
The best data-set we have found, to assess the accuracy of energy demand forecasts, comes from the International Energy Agency’s medium term reports, which have mostly been published annually for oil, coal and sometimes gas, going back to the mid-2000s.
The adjusted error of the estimate for forecasting global oil demand is 0.6% per year. In other words, estimates of global oil demand in N-years time tend to suffer from a forecasting error of N x 0.6%. For example, the IEA’s average estimate for oil demand in 5-years time has tended to be almost 3Mbpd out in either direction.
Methodology. Note that we have excluded the totally anomalous year of 2020 from these calculations, in order to avoid a distorted result.
Overall there has been a small bias towards over-estimating future oil demand. However, this is nuanced. Our data-set covers estimates from the past 15-years, during which time, the world has endured two unexpectedly large economic shocks, one due to the global financial crisis of 2008-09 and the other due to the COVID crisis of 2020-21; and three large price shocks, including the run-up of oil prices in 2007-08, in 2012-14 and in 2021-22.
By contrast, in periods where economic activity was strong and oil prices were low/falling, such as 2010, or 2015-17, oil demand came in 2-3Mbpd higher than had been expected in forecasts from 2-5 years previously.
The data-file also contains data on medium term coal forecasts. Over the past decade, future global coal consumption has consistently been over-estimated, as there was more substitution to use gas and new energies. However, interestingly, 2021-22 appears to be the first year in over a decade where demand is actually higher than forecasts from 3-5 years prior.
Please note, we have not undertaken this analysis to criticize past forecasts. It is difficult to make predictions, especially about the future.
However, we do think that the inherent challenges of forecasting future energy demand in the world is an argument for policymakers to target an energy surplus, with ample spare capacity, in order to avoid the catastrophic impacts of persistent energy shortages. You certainly wouldn’t board a plane whose reserve fuel buffer was lower than the error of the estimate. And once you are in an energy shortage it takes a long time to build new supplies.
How accurate are energy demand forecasts? Please download the data-file to interrogate the numbers that have informed this analysis.